JUST-IN-TIME, LEAN PRODUCTION, AND COMPLEMENTARY PARADIGMS:KANBAN AS A DECENTRALIZED CONTROL SYSTEM FOR JUST-IN-TIME

KANBAN AS A DECENTRALIZED CONTROL SYSTEM FOR JUST-IN-TIME

Prerequisites and Role of Kanban System

As discussed above, continuous improvement, reduced lead times, and a multiskilled workforce are crucial in enabling a production system to respond flexibly to the environmental changes confronting manufacturers. Given these prerequisites and a production schedule that has been leveled, a kanban system can function as a self-regulating, decentralized system for controlling the flow of materials from upstream stages through final assembly.

The Japanese term kanban simply means card or ticket. They are typically enclosed in a protective vinyl cover and contain the following information: part number and name, process name where the kanban is to be used, number of units in the standard container and type of packing, number of kanban cards issued, preceding process outbound stockpoint number, and subsequent process inbound stockpoint number. There is a one-to-one correspondence between the cards themselves and the standard parts containers that they represent. Furthermore, the cards always circulate together with the actual material flow. Through the kanban system, workers understand their operations’ procedures and standards and learn and share the information required for process control. In this way, kanban functions as an information system as well as a means of visual control.

Two kinds of kanbans are used in controlling production and material flows: withdrawal- authorizing kanbans (also called movement kanbans) and production-ordering kanbans. When taken to the preceding process, a withdrawal kanban authorizes the transfer of one standard container of a specific part from the preceding process where it was produced to the subsequent process where it is to be used. A production kanban orders the production of one standard container of a specific part from the preceding process.

Figure 3 illustrates the circulation of kanban cards and containers. Every parts container at an inbound stock point must have a withdrawal kanban attached. When even one of a container’s parts is to be used at the subsequent process, the withdrawal kanban is detached from the container and taken to the outbound stock point of the preceding process together with an available empty container. At the outbound stock point of the preceding process, an already full container of the desired parts is located and its production kanban is removed, with the withdrawal kanban now attached in its place. The full container with withdrawal kanban attached is now ready to be transferred to the inbound stockpoint of the subsequent process, while the empty container is left at the preceding process for later use. In addition, the production kanban that was just removed is now placed in a collection box at the preceding process. These production kanbans are frequently collected and be- come work orders authorizing the production of one more full container of parts. When a new, full container is produced, the production kanban is attached to it and the container is placed in the outbound stockpoint to complete the cycle.

It can be seen that circulation of the kanbans is triggered only by actual usage of parts at the subsequent process. Only what is needed is withdrawn from the preceding process, and then only what is needed for replacement is produced. This chain of usage and production ripples back through upstream stages to the suppliers, with the kanbans functioning as a sort of decentralized, manual

coordination mechanism. Given that major fluctuations in demand have been smoothed and frozen by the leveled production schedule, the kanbans can self-regulate to maintain the needed production quantities while absorbing the inevitable minor variations in everyday operations.

Control Parameters of Kanban System

In considering kanban as a decentralized control system, the following control parameters are nec- essary: number of kanbans in circulation; number of units in the kanban standard container; and kanban delivery cycle a-b-c (where b is number of deliveries per a days and c indicates the delivery delay factor as an indication of replenishment lead time). For example, 1-4-2 means that every 1 day the containers are delivered 4 times and that a new production order would be delivered by the 2nd subsequent delivery (in this case, about a half-day later, given four deliveries per day).

Minimizing work-in-process inventory is a goal of JIT. In keeping with this, the number of units per standard container should be kept as small as possible, with one being the ideal. Furthermore, the number of deliveries per day (b / a) should be set as frequently as possible so as to synchronize with the takt time of the subsequent process, and the delivery delay factor c should be kept as short as possible. Ideally, the number of kanbans in circulation between two adjacent workstations also should be minimized. However, in consideration of practical constraints, a tentative number of kan- bans may be calculated as follows:

This tentative number of kanbans is reconsidered monthly because the daily leveled production requirement may differ under the new monthly production schedule. In addition, the number of kanbans is sometimes further reduced by systematically removing them from circulation in the sys- tem. The resultant reduction of work-in-process inventory will stress the production system and reveal the weakest point for further improvement. Thus, the kanban system is part of the approach used in JIT to move toward the goal of stockless production and achieve continuous improvement of the production system.

Kanban’s Limitations and Alternatives

As previously noted, the kanban system is particularly appropriate for high-volume, repetitive man- ufacturing environments. However, in comparison to the situation when the kanban system was originally created, many industries now face tremendous increases in the variety of products and parts coupled with lower volumes for each individual item. This is seen also in the automobile industry, where more than 50% of parts now have their respective kanbans circulate less than once per day (Kuroiwa 1999). Such a low frequency of circulation leads to undesirably high levels of work-in-process inventory, and the kanban system ends up functioning the same way as the classic double-bin inventory system.

Several alternatives are available when production is not repetitive and / or high volume in nature. For example, when demand is lumpy and cannot be smoothed into a level production schedule, a push method of production coordination may be more appropriate. Karmarkar (1989), for example, discusses this situation and suggests how a hybrid production system utilizing both MRP and kanban may be used. Toyota itself has alternatives to the kanban system (Kuroiwa 1999). For low consump- tion parts, Toyota uses a push method called chakko-hiki (schedule-initiated production). In this approach, production and delivery of the necessary parts is determined in advance, based upon the vehicle final assembly schedule and an explosion of the bills-of-material, rather than through the kanban system. For large components and parts such as engines or seats, yet another system, called junjo-hiki (sequence-synchronized production), is used. Here the sequence of the part’s production and delivery is synchronized item-by-item with the corresponding vehicle sequence of the final as- sembly schedule. For example, the engines are produced and delivered from the engine plant to the final assembly line in the exact order and timing needed for final assembly. In determining which of the various production coordination methods is appropriate, the most important factors for consid- eration are demand volume and delivery lead time. From this standpoint, it should be obvious that kanban is most suitable for parts with relatively high volume and short lead time.

Another important issue in kanban systems is the role of information technology. Kanban is a decentralized, manual system that developed without reliance on computers or other technologies. Given the powerful information technologies presently available, there is some debate over whether to implement electronic kanban systems with the use of bar codes and electronic data interchange (EDI). On the surface, these technologies provide a smart tool in terms of information handling, but it should be noted that the traditional, paper-based kanban has another important role in that workers learn process control and standards through reading the written information on the cards and phys- ically manipulating the system. For this reason, Toyota has limited its use of electronic kanban systems to deliveries involving unavoidably long lead times, such as from distant plants. For example, when a part is consumed at the Kyushu plant, which is located more than 500 kilometers from Toyota’s main plants in the Nagoya region, a withdrawal kanban in bar code form is scanned and its information is transferred through EDI to the preceding supplier or distribution center in the Nagoya region. Then a new withdrawal kanban in bar code form is issued in Nagoya for later attachment and delivery with the new parts, while the old card is disposed of in Kyushu. This is physically analogous to a one-way kanban system, though the electronic information is also utilized for many other purposes, such as physical distribution management, financial clearing systems, and so on.

Case Study of JIT / Kanban Implementation

Aisin Seiki is a Toyota group company that, in addition to manufacturing complex parts for the auto industry, also produces such items as mattresses for sale to retail and institutional customers. As noted in Imai (1997) and Spear and Bowen (1999), Aisin Seiki’s Anjo plant achieved remarkable improvements through the introduction of JIT and kanban production systems. In 1986, prior to beginning JIT, the plant produced 200 styles and variations of mattresses with a daily production volume of 160 units and 30 days worth of finished-goods inventory. Production was based on monthly sales projections and scheduled on a weekly cycle. Due to unreliable forecasts and long lead times, the plant maintained high amounts of inventory yet often experienced material shortages, and sup- pliers had difficulty keeping up with changing orders.

Over the course of several years, the plant introduced various changes to achieve a JIT system with reduced inventories and one-piece production flow. A major change was to produce mattresses only in response to customer demand and based on the sequence in which orders were received. The production schedule cycle was reduced from one week to one day and then eventually to two hours, thereby necessitating many more setups and a reduction in setup times. The quilting machines, for example, now have 60 times as many setup changes. Kanban was introduced for the more popular models so as to produce and hold only the average number of units ordered per day. When a popular model is shipped from the plant to fill an order, its kanban card is returned to the production line as a signal to make a replacement. Lower-volume models are produced only after an order is received. When a large order is received from a hotel, its production is spread out among other orders in order to maintain a level production schedule. As a result of such changes and improvements, the plant is now able to produce a wider variety of end items in higher volumes and with shorter lead times and higher productivity. By 1997, for example, it produced 850 styles of mattresses with a daily produc- tion volume of 550 units and only 1.5 days of finished-goods inventory. At the same time, units produced per person had increased from 8 to 26, and overall productivity had increased by a factor of 2.08.